http://www.westminster.ac.uk/westminsterresearch
Use of Salivary Diurnal Cortisol as an Outcome Measure in
Randomised Controlled Trials: a Systematic Review
Ryan R., Booth S., Spathis A., Mollart S. and Clow A.
This is the publisher version of an article published inAnnals of Behavioral Medicine pp
1-27, First online: 23 March 2016.
Electronic supplementary material: The online version of this
article(doi:10.1007/s12160-015-9753-9) contains supplementary material.
© The Author(s) 2016. This article is published with open access at Springerlink.com
The WestminsterResearch online digital archive at the University of Westminster aims to make the
research output of the University available to a wider audience. Copyright and Moral Rights remain
with the authors and/or copyright owners.
Whilst further distribution of specific materials from within this archive is forbidden, you may freely
distribute the URL of WestminsterResearch: ((http://westminsterresearch.wmin.ac.uk/).
ORIGINAL ARTICLE
Use of Salivary Diurnal Cortisol as an Outcome Measure
in Randomised Controlled Trials: a Systematic Review
Richella Ryan, Dr.
1,2
&
Sara Booth, Dr.
1,2
&
Anna Spathis, Dr.
1
&
Sarah Mollart, Dr.
3
&
Angela Clow, Prof.
4
# The Author(s) 2016. This article is published with open access at Springerlink.com
Abstract
Background Dysregulation of the
hypothalamic-pituitary-adrenal (HPA) axis is associated with diverse adverse health
out-comes, making it an important therapeutic target. Measurement
of the diurnal rhythm of cortisol secretion provides a window into
this system. At present, no guidelines exist for the optimal use of
this biomarker within randomised controlled trials (RCTs).
Purpose The aim of this study is to describe the ways in
which salivary diurnal cortisol has been measured within
RCTs of health or behavioural interventions in adults.
Methods Six electronic databases (up to May 21, 2015) were
systematically searched for RCTs which used salivary diurnal
cortisol as an outcome measure to evaluate health or
behav-ioural interventions in adults. A narrative synthesis was
un-dertaken of the findings in relation to salivary cortisol
meth-odology and outcomes.
Results From 78 studies that fulfilled the inclusion criteria, 30
included healthy participants (38.5 %), 27 included patients
with physical disease (34.6 %) and 21 included patients with
psychiatric disease (26.9 %). Psychological therapies were most
commonly evaluated (n=33, 42.3 %). There was substantial
heterogeneity across studies in relation to saliva collection
pro-tocols and reported cortisol parameters. Only 39 studies (50 %)
calculated a rhythm parameter such as the diurnal slope or the
cortisol awakening response (CAR). Patterns of change in
cor-tisol parameters were inconsistent both within and across studies
and there was low agreement with clinical findings.
Conclusions Salivary diurnal cortisol is measured
inconsis-tently across RCTs, which is limiting the interpretation of
findings within and across studies. This indicates a need for
more validation work, along with consensus guidelines.
Keywords Salivary cortisol . Diurnal rhythm . Randomized
controlled trial . Systematic review . Intervention
Introduction
The hypothalamic-pituitary-adrenal (HPA) axis is known to
be an important pathway in the regulation of the physiological
stress response. HPA axis dysregulation has been shown to be
Electronic supplementary material The online version of this article
(doi:10.1007/s12160-015-9753-9) contains supplementary material,
which is available to authorized users.
* Richella Ryan
richella.ryan@nhs.net
Sara Booth
sb628@cam.ac.uk
Anna Spathis
anna.spathis@addenbrookes.nhs.uk
Sarah Mollart
sarahmollart@nhs.net
Angela Clow
clowa@westminster.ac.uk
1
Palliative Care Department, Cambridge University Hospitals NHS
Foundation Trust, Elsworth House, Box 63, Hill
’s Road,
Cambridge CB2 0QQ, UK
2
Department of Oncology, University of Cambridge, Hutchison/MRC
Research Centre, Cambridge Biomedical Campus, Box 197,
Cambridge CB2 0XZ, UK
3
St. Nicholas Hospice Care, Hardwick Lane, Bury St.
Edmunds, Suffolk IP33 2QY, UK
4
Department of Psychology, University of Westminster, 101 New
Cavendish Street, London W1W 6XH, UK
DOI 10.1007/s12160-015-9753-9
associated with important health outcomes including
psychi-atric illness [
1
], cardiovascular mortality [
2
], cancer prognosis
[
3
,
4
], and frailty and cognitive decline [
5
]. These associations
are thought to be mediated by the deleterious effects of
chron-ic stress on HPA axis function [
6
], with secondary effects on
metabolic, immune and psychobiological systems [
7
]. The
many associations between HPA axis dysregulation and
markers of health status suggest that HPA axis modulation
by therapeutic interventions may have a role in disease
treat-ment and prevention. In order to demonstrate this, accurate
and feasible measurement of HPA axis function within
randomised controlled trials (RCTs) is necessary.
The use of salivary cortisol as a biomarker of stress and
HPA axis function is a well-established practice in stress
re-search, dating back to at least 20 years [
8
]. Due to marked
diurnal variation in cortisol hormone secretion throughout the
day [
9
], a variety of methods of salivary cortisol collection and
analysis have been explored and utilised in an attempt to
iden-tify the most representative summary measure of HPA axis
function. Essentially, two broad approaches have been taken
[
9
]. The first approach is to measure HPA axis reactivity to a
standardised acute stressor. Whilst this approach is useful,
in-terpretation of the results is limited by the need to consider the
time of day the stressor is administered, as well as the nature of
the stressor. The second approach is to measure basal or
unstimulated HPA axis function; thus avoiding the need to
administer a stressor.
Measurement of basal HPA axis function has evolved
con-siderably over the past two decades, as theoretical and
empir-ical knowledge regarding the stress system, cortisol
measure-ment and disease associations has increased. In the early days
of salivary cortisol research, a single salivary cortisol measure,
collected at a pre-specified time, was used to estimate basal
HPA axis function, but this methodology proved to be
unreli-able, with large intra-individual and inter-individual variation
[
10
]. Another common approach was to measure average or
total cortisol exposure over a 12–24-h period [
11
]. Whilst this
approach provides a summary measure, it does not
accommo-date the complex nature of HPA axis aberration, with both
hypocortisolism and hypercortisolism now recognised to be
linked to chronic stress [
12
]. Increasingly, therefore, there has
been a move towards measuring the circadian rhythm of
diur-nal cortisol secretion, rather than focusing on absolute cortisol
concentration [
11
].
Typically, under basal conditions, a healthy HPA axis is
characterised by a distinctive circadian pattern of cortisol
se-cretion, whereby cortisol rises to a peak within 30
–45 min of
waking and then falls to a nadir during sleep at approximately
midnight [
9
,
11
]. The major measurable parameters of this
diurnal rhythm are (1) the cortisol awakening response
(CAR), which is the rise in cortisol during the first 30–
45 min following awakening [
9
], and (2) the diurnal cortisol
slope, which is the rate of decline in cortisol levels across the
day, from morning to evening [
11
]. This normal rhythm
be-comes disrupted when the HPA axis bebe-comes dysregulated
[
9
,
13
], the pattern of disruption varying depending on the
context or condition studied. In general, an abnormal cortisol
awakening response, both abnormally large and small, or a
flattened diurnal cortisol slope appear to be consistent markers
of HPA axis dysfunction [
11
]. Importantly, there is evidence
that these parameters are independently regulated, with the
cortisol awakening response being mediated by an
extra-pituitary pathway to the adrenal from the suprachiasmatic
nu-cleus [
14
]. Thus, these parameters are believed to represent
different aspects of HPA axis function [
14
–
16
].
Due to the many possible ways of measuring salivary
cor-tisol as a biomarker in stress research, it is necessary to reach
consensus regarding the most appropriate methodology, so
that the results of different research studies can be compared
and so as to avoid waste in the design, conduct and reporting
of studies. Adam and Kumari [
11
] have reviewed the use of
salivary diurnal cortisol in epidemiological studies and have,
accordingly, published recommendations. They found that the
cortisol awakening response, the cortisol slope and the area
under the daytime cortisol curve (AUC) were most commonly
measured within large epidemiological studies and have been
most robustly linked with psychosocial phenomena and health
outcomes, implying clinical relevance. They recommend that
these parameters should each be assessed as separate
indica-tors of HPA axis function and that the cortisol collection
schedule be sufficient to measure, at minimum, the cortisol
awakening response and the diurnal slope over more than
1 day.
There is no guidance available for the measurement of
salivary diurnal cortisol within interventional studies, and
lit-tle is known about how salivary cortisol has been employed,
to date, as a biomarker within RCTs of health and behavioural
interventions. The inherent complexity of salivary diurnal
cor-tisol as a biomarker is likely to pose particular challenges
within RCTs. Given that the diurnal profile is essentially a
composite of at least three measurement parameters (the
cor-tisol awakening response, diurnal slope and area under the
curve), each reflecting different aspects of HPA axis function,
it is possible that experimental interventions will have
differ-ent effects on differdiffer-ent parameters. This is likely to impact on
a priori decisions about the primary measurement parameter
of interest, on hypotheses about directions of change and on
conclusions about efficacy, target engagement and
mecha-nisms of action. As well as these challenges, there are
con-cerns in the literature about the long-term stability of this
biomarker over periods of greater than 1 month [
17
], as well
as concerns about its reliability in the shorter term due to
day-to-day state effects [
18
] and the effects of non-compliance
[
19
]. Concerns have also been raised about the responsiveness
of the biomarker and how different contexts and populations
To assess whether specific guidance is necessary, we
sys-tematically reviewed the literature with the aim of describing
the RCTs of health and behavioural interventions which have
used salivary diurnal cortisol as an outcome measure,
partic-ularly focusing on salivary diurnal cortisol methodology and
findings. Specifically, we aimed to explore the following
questions:
1. Which health and behavioural interventions have been
evaluated using salivary diurnal cortisol?
2. What populations have been evaluated?
3. What collection protocols have been used to obtain a
di-urnal cortisol profile?
4. What parameters of the diurnal cortisol profile have been
measured?
5. Where a change in a cortisol profile parameter is
ob-served, when in the follow-up period does it occur?
6. How often is there consistency between the clinical and
cortisol response to the intervention?
Methods
The protocol for this review is available in the Electronic
Supplementary Material
1
.
Study Inclusion and Exclusion Criteria
We restricted the sample to RCTs only. Though the review
question is relevant to non-randomised uncontrolled
longitu-dinal studies also, we chose to select RCTs only in order to
reduce the scope of the search and the heterogeneity of the
sample; this was deemed necessary given the broad review
question with respect to interventions and sample populations.
We also expected that RCTs would be of higher quality than
other study designs, thus providing more reliable information.
We defined
‘diurnal cortisol profile measurement’ as the
col-lection of at least two samples of salivary cortisol over at least
1 day. This would enable the calculation of the diurnal slope or
the cortisol awakening response, at minimum. Within the
con-text of an RCT, there needed to be evidence that this
measure-ment had been obtained on at least one occasion before the
intervention and on at least one separate occasion (i.e. a
sep-arate day or period of days) after the intervention. Using these
definitions, we adhered to the following inclusion and
exclu-sion criteria:
Inclusion Criteria
1. Population: any adult (>18 years) population.
2. Study design: randomised controlled trials.
3. Interventions: any type of therapeutic intervention
de-signed to improve an aspect of health or well-being,
ex-cluding exogenous corticosteroids.
4. Control or comparator: any type of control or comparator.
5. Outcome measures: studies that use salivary diurnal
cor-tisol profile measurement as a primary or secondary
out-come measure.
Exclusion Criteria
1. Non-RCT studies, including quasi-randomised
con-trolled trials and trial protocol reports without results
2.
Studies which use non-diurnal salivary sampling
methods e.g. single salivary cortisol measures pre and
post an intervention or salivary cortisol pre and post a
stress-task
3. Studies which evaluate the effects of exogenous
gluco-corticoids (any type or route) on salivary cortisol
4. Studies which evaluate the cortisol response to
stress-inducing interventions or conditions
5. Studies which measure diurnal cortisol under
laboratory-induced conditions (e.g. light-wake conditions)
6. Studies in which the diurnal profile is not measured both
before and after the intervention
7. Studies in which the diurnal profile is obtained on the
same day as the intervention with a view to assessing its
acute effects within the day
8. Studies in people with Cushing
’s disease
9. Animal studies
10. Abstract publication available only
11. Dissertation or non-journal publication available only
12. Non-English language publications
Search Methods for Identification of Studies
On 21 May 2015, we searched the following electronic
data-bases using NHS Evidence Healthcare Datadata-bases Advanced
Search tool: MEDLINE (1980 to May 2015), CINAHL (1980
to May 2015), PsychINFO (1806 to May 2015), AMED
(1985 to May 2015), EMBASE (1974 to May 2015) and the
Cochrane Central Register of Controlled Trials. We sought to
identify a combination of keywords and MESH terms in the
titles and abstracts of papers, adapting the search strategy, as
appropriate, for each database. By way of example, the
fol-lowing keywords, MESH terms and publication types were
searched in MEDLINE: [(‘cortisol’ AND ‘saliva*’) OR
(HY-DROCORTISONE/ AND SALIVA/)] AND [(
‘randomized
controlled trial’ OR ‘controlled clinical trial’ OR
‘random-ized’ OR ‘placebo’ OR ‘randomly’ or ‘trial’ OR ‘groups’,
more detailed search strategies, including strategies used in
the other databases.
Study Selection for Inclusion in the Review
All abstracts generated from the electronic searches were
exported to Endnote X3 for removal of duplicates. One review
author (RR) screened the titles and abstracts for the eligibility
criteria, and where eligibility could not be determined, the
full-text article was obtained. All full-text articles were
reviewed for eligibility by RR, and a random selection of these
articles (37 %) were reviewed independently by the other four
authors (SB, AS, AC and SM), each reviewing a different
selection, to ensure that the eligibility criteria were being
cor-rectly interpreted and adhered to. The inclusion criteria were
applied in a hierarchical manner, first checking the population,
then the study design, then the intervention and finally the
cortisol methodology. Any disagreement was discussed in
the first instance between the two authors in question. If
con-sensus was not achieved between the two authors in question,
a third party (one of the other authors) was consulted.
Data Collection and Extraction
All full-text articles which were deemed eligible were
reviewed in more detail for data extraction. RR completed
the data extraction for all eligible texts. In addition, another
author independently completed data extraction on 10 % of
the eligible articles to ensure that data were being extracted
appropriately. A data extraction form containing the following
fields was used to summarise the pertinent details of the study:
trial ID, eligibility criteria checklist with decision outcome,
study design, population, intervention/control, salivary
corti-sol collection protocol details, other outcome measures,
sali-vary cortisol analysis details, salisali-vary cortisol results and
clin-ical outcome results.
Assessment of Quality and Relevance
The quality of each study and its relevance to the review aim
were assessed using the Gough Weight of Evidence
frame-work [
21
], which uses four domains of assessment (A, B, C
and D), rating each domain as low, moderate or high. As the
overarching aim of this review was to describe salivary
corti-sol methodology and findings in RCTs, the relevance of each
study was assessed purely in relation to the degree to which it
contributed information towards this aim.
Within the first domain of this framework, judgments are
made in relation to the generic quality of execution of the
study independent of the review aim (Weight of Evidence
A). Within the second and third domains, judgements are
made in relation to the specific aim of the review, including
the appropriateness of the study design to the review aim
(Weight of Evidence B) and the focus of the study, including
its objectives and reporting, relative to the review aim (Weight
of Evidence C). An overall judgement of quality and
rele-vance (low, moderate or high) is formed by combining the
assessments for these domains (Weight of Evidence D). This
framework is weighted more heavily towards relevance than
quality and was chosen as a means of highlighting those
stud-ies which provided the most relevant information towards the
review aim. This was deemed to be the most appropriate
ap-proach to appraising the literature included in this review,
given that we were not concerned with evaluating the efficacy
of specific interventions.
Data Synthesis and Presentation
The selection process is presented using a PRISMA flow
chart. A narrative synthesis of the scope, characteristics and
findings of the selected studies is given and presented in
ta-bles. To enhance clarity and facilitate comparison, individual
studies were organised into four categories according to the
intervention being evaluated, and a separate table of studies
was created for each category. Counts and percentages were
used to describe data patterns across all studies and between
study intervention categories. Medians and interquartile
ranges were calculated to describe key features of the salivary
collection protocol across studies and to describe the
frequen-cy and distribution of the follow-up time-points at which
cor-tisol findings occurred. For a corcor-tisol parameter, a significant
finding was considered to be present if a study reported a
statistically significant within- or between-group effect from
baseline to follow-up, for either the intervention or the
com-parator group. If a significant finding was also reported for at
least one clinical outcome measure in the same study, using
the same statistical tests, the cortisol findings were considered
to support the clinical outcome. In the same way, if there was
no evidence of a statistically significant effect for both the
cortisol and clinical outcomes, cortisol findings were
consid-ered to support clinical findings.
Results
Selection Process
The process of screening and reviewing articles for eligibility
is summarised in Fig.
1
. The database search identified 2374
articles. After removal of duplicates using Endnote X3, 1812
potentially relevant abstracts were identified. Screening of
these abstracts led to the selection of 219 full-text articles for
more detailed eligibility assessment. Of these full-text articles,
78 studies were selected for inclusion in the review after
re-moval of ineligible studies and after exclusion of duplicate
reports of the same study. The most common reason for
exclusion of articles related to salivary cortisol methodology.
After excluding studies due to ineligible populations,
inter-ventions and designs, 87 of the remaining 175 RCTs (50 %)
were excluded because the salivary cortisol measurements
therein did not allow analysis of the diurnal rhythm. In most
cases, this was due to the measurement of cortisol on a single
occasion before and after the intervention being evaluated. A
small number of RCTs (n=10), which did include diurnal
cortisol measurements, were later excluded from the review
because their cortisol findings were not adequately reported or
because their measurements were conducted in a way that was
not comparable with the other studies.
Characteristics of Included Studies
Included studies were published from 2003 to May 2015.
There has been a notable increase in the number of published
RCTs using salivary diurnal cortisol as an outcome measure in
the past decade, with the yearly rate increasing from 2 studies
per year in 2003 and 2004 to 14 studies in 2013 and 11 studies
in 2014 (see Fig.
2
). Indeed, over 50 % of the included studies
have been published since 2012. Pertinent characteristics of
individual studies are presented within Tables
1
,
2
,
3
and
4
,
with each table representing one of four study categories and
each study being organised into one of such categories
accord-ing to the intervention beaccord-ing evaluated: (1) RCTs evaluataccord-ing
psychosocial interventions, (2) RCTs evaluating
pharmaco-logical (including nutritional) interventions, (3) RCTs
evalu-ating complementary therapies and (4) RCTs evaluevalu-ating all
other types of interventions.
Most commonly, studies evaluated psychosocial
interven-tions (n=33, 42.3 %), such as cognitive behavioural therapy,
mindfulness and psychotherapy (see Table
1
).
Pharmacologi-cal therapies, including nutritional therapies, comprised the
sec-ond most common intervention category (n=22, 28.2 %),
within which eight studies evaluated anti-depressant
medica-tions (see Table
2
). Complementary therapies were evaluated
in 14 studies (17.9 %), with six of these studies evaluating
Ident
ification
Records identified through database
searching
(n =2,374)
Records after duplicates removed
(n =1812)
Screening
Records excluded
(n =1,593)
Records screened
(n =1812)
Full-text articles excluded
(n =141)
-Two different papers reporting the same
study= 10
-Not available in English=5
-Unable to access full-text article=1
-Population < 18 yrs= 2
-Non-therapeutic intervention (intervention
used as a ‘stressor’ or experimental
‘condition’)= 3
-Not an RCT=23
-Diurnal cortisol profile not measured= 87
-Diurnal cortisol profile not measured both
before and after the intervention or
measured on the day of the intervention to
assess acute effects=8
-Cortisol results not reported in the text=2
Full-text articles assessed
for eligibility
(n =219)
Eligibility
Included
Studies included in
narrative synthesis
(n=78)
Fig. 1 PRISMA flow diagram
illustrating the identification of
studies
yoga (see Table
3
). Nine studies evaluated treatments which
did not fall into the three major categories and were, therefore,
classified as
‘other’ (see Table
4
). These treatments included
exercise (n=3), cranial electrostimulation (n=2), a lifestyle
intervention (n=1), dietary restriction (n=1), prayer (n=1)
and light treatment (n=1).
Interventions were evaluated in a wide variety of study
populations. Overall, across all studies, these populations
could be broadly classified as follows: people who were
healthy or at risk of disease (n=30, 38.5 %), patients with
physical or psychosomatic disease (n=27, 34.6 %) and
pa-tients with a psychiatric diagnosis (n=21, 26.9 %). Physical
and psychosomatic disease categories included current or
pri-or cancer (n=14; predominantly breast cancer), cardiovascular
disease or metabolic syndrome (n=3), human
immunodefi-ciency virus (n= 3), dementia (n =3), Parkinson’s disease
(n = 1), irritable bowel syndrome (n = 1), tension headache
(n = 1) and fibromyalgia (n = 1). Psychiatric pathology
in-cluded depression (n = 12), anxiety disorder (n = 2),
co-morbid anxiety and depression (n = 1), post-traumatic
stress disorder (n = 3), adjustment disorder (n = 1), bipolar
disorder (n = 1) and alcoholism (n = 1). Within study
cate-gories, psychosocial intervention studies and
pharmaco-logical studies most commonly evaluated people who
were healthy or at risk of disease (42 and 54.5 % of
psychosocial and pharmacological studies, respectively),
whereas complementary therapy studies most commonly
evaluated patients with physical or psychosomatic disease
(86 % of complementary therapy studies).
Overall, across study categories, the median study sample
size was 55 participants (IQR 34–77), with the smallest study
including only 12 participants [
76
] and the largest including
379 participants [
57
]. Median sample size was similar
be-tween study categories: 59 (IQR 34–74) for psychosocial
in-terventions, 51 (IQR 41–76) for pharmacological
interven-tions, 49 (IQR 21–90) for complementary therapies and 59
(IQR 24–87) for other interventions. Eight studies were
re-ported as pilot, feasibility or exploratory studies. Overall, the
median length of follow-up from baseline was 10 weeks (IQR
4 to 21) and ranged from 1 to 72 weeks.
Quality and relevance (aggregate score) were rated as high
for 20 studies and moderate for 58 studies. No study was
given an aggregate score of low, reflecting the exclusion of
studies of low relevance by the eligibility criteria. Of note,
many studies were of low quality with respect to their RCT
design but of high relevance with respect to the review aim,
resulting in a high overall aggregate score using Gough’s
framework [
21
]. See
Appendix B
for the breakdown of scores
per domain for each included study.
Salivary Cortisol Collection and Analysis Methodology
Pertinent details relating to the salivary cortisol collection
pro-tocols and parameters used in individual studies are presented
in Tables
1
,
2
,
3
and
4
.
In relation to saliva collection, the median number of days of
saliva collection per time-point across studies was 1 day (IQR 1–
2). As the median suggests, the majority of studies (n=57,
Fig. 2 Number of eligible RCTs
published per year. The year 2015
was excluded from this graph as
the complete results for this year
are not yet available
Table 1
Randomised controlled trials evaluating psychosocial interventions: study characteristics, salivary cortisol methodology and main findings
Study ID Main study objective(s) StudyPopulationc Intervention Comparator or Control Main outcome measures Assess-ment
period Saliva collection protocola Diurnal cortisol parameters analysedd
Main cortisol findings Main clinical findings Quality/ Relevance b Bergen-Cico et al. 2014 (22)
To assess the efficacy of a brief primary care mindfulness programme in military veterans with post-traumatic stress disorder (PTSD) by examining changes in cortisol as a biomarker. 40 military veterans Primary-care-based brief mindfulness-based stress reduction programme (PCbMP) delivered in weekly 90 min sessions over 4 wks. Home practice encouraged. Primary Care treatment as usual; they were given the option of receiving PCbMP after 12 wks. 1) Salivary cortisol 2) Psychometric measures for depression and PTSD. 4 wks 2 day 5 sample points (on awakening, 45 min after awakening and 3 other points)
1) Total AUC with respect to ground (AUCg) 2) AUC with respect to increase (AUCi) 3) CAR calculated as the AUC from awakening to 45 min
There was a significant group x time interaction effect on AUCi, with AUCi decreasing in the intervention group and increasing in the control group over time. Pre-post within group analysis indicated that the CAR reduced significantly in the intervention group but not in the control group.
There was a significant negative correlation between changes in AUCi and changes in depression score over time for the group as a whole but no correlation between the changes in cortisol and psychometric variables when each group was analysed separately.
High
Bormann et al. 2009 (23)
To determine the effect of a spiritually-based mantram intervention, in comparison with a control, on average daily salivary cortisol levels in HIV-infected adults. 71 HIV infected adults Spiritually-based mantram intervention, delivered over 5 weeks in the form of face-to-face classes Attention-based control intervention delivered in the format 1) Spiritual wellbeing scale, with faith/assurance subscale as a primary outcome 2) Salivary cortisol 10 wks 1 day 4 sample points (7am, 11am, 4 pm, and 9pm) Mean diurnal cortisol level
There was no significant difference in mean daily cortisol level between groups at 5 and 10 weeks
Faith/assurance levels were significantly higher in the intervention group in comparison with the control group at 5 and 10 weeks Moderate Bougea et al. 2013 (24) To evaluate whether emotional freedom technique affects the frequency and intensity of headaches in adults suffering from tension-type headaches 35 adults suffering from frequent tension-type headaches Emotional freedom technique practised twice daily over 8 weeks Standard care (continued use of medication) 1) Frequency and intensity of headaches 2) Perceived stress 3) Global health and health control measures 4) Sleep parameters 5) Salivary cortisol 8 wks 1 day 2 sample points (8am and 8pm) Time-specific cortisol level
There were no significant differences in cortisol levels (8am or 8pm) between the intervention and control groups at 8 weeks
Headache frequency and intensity, and perceived stress, were significantly lower in the intervention group in comparison with the control at 8 weeks
Moderate
Carlson et al. 2013 (25)
To compare the efficacy of mindfulness-based cancer recovery (MBCR) and supportive-expressive group therapy (SET) in distressed survivors of breast cancer 271 distressed survivors of stage 1-III breast cancer MBCR delivered in 90 min weekly group sessions over 8 wks plus a 6-hour workshop between weeks 6 and 7 SET delivered in 90 min weekly sessions over 12 wks OR Control condition: 1-day (6-hour) didactic stress management 1) Self-reported mood (primary) 2) Salivary cortisol (primary) 3) Self-reported stress 4) Quality of life 5) Self reported social support Up to 12 wks 3 days 4 sample points (On awakening, noon, 5pm, bedtime) 1) Diurnal slope 2) Time-specific cortisol level
There was a significant group x time interaction effect for diurnal slope, with slopes becoming steeper after SET and MBCR, compared with the control (SMS). In the within-group pre-post analysis, there was no significant change in slope in either MBCR or SET, however, but the slope
There were significant group x time interaction effects for mood and stress in the intention to treat analysis. Follow-up pair-wise comparisons for each group indicated no significant differences between the groups for mood disturbance,
High
seminar (SMS) became significantly flatter in the control (SMS). There was a significant difference in the change in bedtime cortisol between the MBCR group and the control group, with an increase in the SMS group and a slight decrease in the MBCR group.
however. Stress symptoms, however, reduced significantly in the MBCR group in comparison with both the SET and the SMS groups.
Cash et al. 2015 (26)
To evaluate the effects of mindfulness-based stress reduction (MBSR) on symptoms and neuroendocrine function in fibromyalgia 91 females with fibromyalgia MBSR delivered in weekly 2.5hour sessions over 8 wks Waiting list group x 8 wks 1) Perceived stress 2) Pain 3) Sleep quality 4) Fatigue 5) Fibromyalgia impact scale 6) Salivary cortisol 16 wks 2 days 6 sample points (On awakening, 45 min after awakening, noon, 4pm, 8pm, bedtime) 1) Diurnal slope (excluding +45min sample) 2) Mean waking level 3) CAR-calculated as the mean percent increase from waking to +45 min 4) CAR slope
There were no significant differences between the two groups at 8 wks or 16 wks for any of the cortisol parameters, controlling for baseline values.
There were significant reductions in perceived stress, sleep problems, fatigue and symptom severity in the MBSR group relative to the control at 8 wks, but not in pain or physical functioning. The effects on fatigue were not maintained at 16 wks. High Chan et al. 2006 (27) To investigate the psychophysiological effects of different psychosocial intervention methods for breast cancer patients 76 patients with breast cancer (stage I-III) Body-mind-soul intervention OR Supportive expressive therapy OR Social support Non-intervention control 1) Psychological measures including stress 2) Salivary cortisol 32 wks 2 days 5 sample points (On awakening, 45 min after awakening, 12pm, 5pm and 9pm) 1) AUC 2) Diurnal slope
There was a significant reduction in cortisol AUC in the body-mind-soul group, but not the other groups, at 8 months
There was a significant improvement in a range of psychological measures in the body-mind-soul group only.
Moderate Delle Chiaie et al. 2012 (28) To investigate if group psychoeducation normalizes HPA axis function in patients with bipolar disorder (preliminary results of an ongoing investigation) 20 patients with stabilized bipolar disorder Group psycho-education delivered in 21 sessions. Continuation of treatment as usual (TAU) along with 21 weekly group meetings in which no special instruction was given 1) Depression 2) Mania 3) Compliance with drug treatment 4) Salivary cortisol 20 wks 1 day 5 sample points (08.00am, 30 min and 60 min after, 1pm, 8pm)
The first 3 sample points (representing the CAR) and the last 2 sample points (diurnal decline) were analysed in separate ANOVA models at each timepoint, with sample time as the within group factor and treatment as the between group factor.
There were no significant differences between groups at baseline. After treatment, there was a significant difference between groups in the morning pattern of cortisol secretion (CAR), with an increase in the CAR in the psychoeducation group.
There were no significant changes in the clinical measures
Moderate
Gaab et al. 2006 (30)
To evaluate the effects of cognitive behavioural stress management (CBSM) on psychological and somatic wellbeing and neuroendocrine responses in students preparing for an academic exam. 28 healthy economics students Cognitive behavioural stress management programme over 4 weeks Waiting list control 1) Psychological measures including anxiety, stress and depression 2) Somatic symptoms 3) Salivary cortisol 4 wks 1 day 9 sample points (On awakening, 15, 30, 45 and 60 min AND 8am, 11am, 3pm and 8pm) 1) CAR measured by AUC over the first hour 2) Total AUC over full day
There was a significant time x group interaction effect for the CAR, with a significant change in the CAR in the intervention group
There was a significant time x group interaction effect on anxiety and somatic symptoms, with less anxiety and somatic symptoms in the intervention group.
High Feicht et al. 2013 (29) To explore whether ‘happiness training’ improves individual happiness and satisfaction with life and relieves stress in an occupational setting. 147 adult volunteers working in an insurance company Web-based happiness training for 7 wks Waiting list group 1) Happiness and satisfaction with life on a visual analogue scale 2) Wellbeing 3) Stress warning signals 4) Mindfulness 5) Recovery experience 6) Flourishing 7) Salivary cortisol and alpha-amylase (subsample only- n=45) 8) Attention network test 11 wks 1 day 3 sample points (on awakening, 30 min later, 8pm) 1) CAR: 30 min sample minus awakening sample 2) ‘Morning activity’: mean of 30 min sample plus awakening sample 3) ‘Evening activity’: 8pm level 4) Circadian amplitude: evening sample minus awakening sample
There were no significant within or between group differences for any of the cortisol parameters
There were significant within and between group changes for a range of psychological measures, with evidence of increased happiness and satisfaction in life and reduced stress in the intervention group.
Table 1
(continued)
Gex-Fabry et al. 2012 (31) 1) To confirm the efficacy of mindfulness-based cognitive therapy (MBCT) compared with Treatment as Usual (TAU) in reducing relapse risk in patients in remission from recurrent depression 2) To examine possible changes in diurnal salivary cortisol profiles over the follow-up period60 adult patients in remission from recurrent (≥3 episodes) depression, off anti-depressant medication for at least 3 months MBCT plus TAU MBCT was delivered in 8 weekly sessions over 2 months TAU Any care deemed necessary 1) Rate of relapse of depression 2) Time to relapse of depression 3) Salivary cortisol 56 wks 1 day 7 sample points (On awakening, followed by 15, 30, 45 and 60 min post-awakening, and at 3pm and 8pm) 1) CAR: AUC above the minimum concentration in the first hour after awakening 2) Diurnal Slope: difference between wake-up and last evening values, divided by the time interval between these samples 3) Total AUC
There were no significant time x group effects for any cortisol parameter measured, indicating that there was no significant difference between groups at any follow-up point in relation to cortisol
There was no significant difference in depression relapse rate between the two groups over the follow-up period High Holt-Lunstad et al. 2008 (32)
To test the hypothesis that couple-based emotional support training that enhances warm physical contact between marital partners may induce increases in oxytocin 34 healthy married couples ‘’Warm Touch” Support Enhancement Intervention- training provided over 4 weeks Couples were advised to keep a diary of physical affection and mood over 4 weeks 1) Ambulatory blood pressure 2) Plasma and salivary oxytocin 3) Salivary cortisol 4) Salivary alpha-4 wks 1 day 5 sample points (On awakening, 7am, 12pm, 5pm, 10pm)
AUC Mixed model analysis indicated that there was no main effect of the intervention on salivary cortisol AUC
The intervention had a significant effect on salivary oxytocin and alpha amylase levels but no significant effect on plasma oxytocin levels or
Moderate
activity and decreases in stress hormones and blood pressure.
amylase ambulatory blood pressure. Hsiao et al. 2011 (33) To examine the psychobiological effects of psychotherapy coupled with pharmacotherapy versus pharmacotherapy alone in patients with major depressive disorder
63 adults (18-65 years) with major depressive disorder attending an outpatient psychiatry clinic Combined therapy (COMB): mind-body-spirit psychotherapy delivered over 8 weeks in addition to pharmacotherapy Monotherapy (MT): pharmacotherap y alone 1) Depression scale 2) Anxiety scale 3) Salivary cortisol 32 wks 1 day 5 sample points (On waking, 45 min after waking, 12 noon, 5pm, 9pm) 1) Time-specific cortisol level 2) Diurnal pattern- modelled using mixed modelling techniques
Mixed model analysis indicated that there was no significant time x group interaction effect on second-order polynomial diurnal cortisol change pattern (planned analysis) but there was a significant time x group interaction effect on first-order polynomial diurnal cortisol change pattern (post-hoc analysis). Analysis of the change in cortisol slope from baseline to each follow-up timepoint indicated that the change in the cortisol slope over the 3 follow-up points differed significantly between COMB and MT, with the slope becoming steeper in the COMB group.
There were no time x group interaction effects or time main effects on depression scores or state anxiety scores. However, there was a group main effect in change in anxiety scores, with a greater change observed in the COMB group. Moderate Hsiao et al. 2012 (34)
To understand the effect of a body-mind-spirit intervention on depression state, sense of meaning in life, and diurnal cortisol patterns in breast cancer survivors
48 breast cancer survivors attending a surgical outpatient department Body-mind-spirit group therapy delivered for 2 hours weekly over 2 months Educational session delivered as 1 session 1) Depression state measure 2) Meaning in life measure 3) Salivary cortisol 32 wks 1 day 6 sample points (On awakening, 30 min, 45 min, 12 noon, 5pm, 9pm) 1) Diurnal decline: calculated based on regression of the 6 cortisol levels 2) Time-specific cortisol level
The control group developed a significantly higher 21.00 cortisol level and a flatter slope, in comparison with the intervention, over the 8 month follow-up period
There were no significant differences between groups over time in relation to the depression scores and the ‘Meaning-in-life Questionnaire-Presence scores’. However, there was a significant difference between groups in the ‘Meaning in Life Questionnaire- Search scores’ over 5 months (but not 8 months)
Moderate
Jensen et al. 2012 (36)
To test the effects of mindfulness-based stress reduction (MBSR) on attention in healthy volunteers 48 healthy volunteers who were meditation novices, recruited from a university. Mindfulness-based stress reduction (MBSR), delivered once weekly in 2.5 hour sessions, over 8 wks. It also included home assignments and an intensive retreat. Non- mindfulness-based stress reduction (NMSR): resembled MBSR but did not include meditation practices nor training in a non-judgmental attitude OR Non-treatment inactive control 1) 5 attentional tasks 2) Salivary cortisol 3) Perceived stress 4) Mindfulness and awareness 8 wks 1 day 5 sample points (on awakening, 15min,30min, 45min and 60 min after awakening)
1) CAR AUCg: area under the curve with respect to ground, reflecting magnitude 2) CAR AUCi: area under the curve with respect to increase, reflecting pattern
There was a significant within-group reduction in AUCi in the MBSR group with no significant within-group changes in the other groups.
There was a group x time interaction effect for AUCg when MBSR was compared to the control but not when MBSR was compared with NMSR.
There was a significant improvement in mindfulness and perceived stress in the MBSR group, with no significant change in the other groups. There was a group x time interaction effect for these measures when MBSR was compared to the control but not when MBSR was compared to NMSR. Attentional measures not relevant to this review. Moderate Klatt et al. 2009 (37) To determine whether low-dose mindfulness-based stress reduction (MBSR-ld) significantly decreases symptoms of stress and to yield adherence rates similar to those for traditional MBSR interventions 48 healthy university staff members aged 18-60 years MBSR-ld delivered over 6 weeks Waiting list control over 6 weeks 1) Perceived stress 2) Sleep 3) Level of mindfulness 4) Salivary Cortisol 6 wks 2 days 3 sample points (20 min after waking, 1pm and 10pm) Mean diurnal cortisol level
There was no significant change in cortisol levels over time in either group.
There was a significant improvement in mindfulness and perceived stress in the intervention group, but not in the control group.
Moderate Hsiao et
al. 2014 (35)
To examine the effects of psychotherapy on depressive and anxiety symptoms, suicidal ideation and diurnal
71 patients with adjustment disorder associated Body-mind-spirit (BMS) psychotherapy received in addition to Control: similar to treatment as usual which includes medication and 1) Depression 2) Anxiety 3) Occurrence of suicidal ideation 4) Salivary 56 wks 1 day 6 sample points (on awakening, 30 min after, Diurnal slope calculated by regressing cortisol values at all 6 sample points
There was a significant group x time interaction effect for diurnal slope, with a significantly steeper slope at 56 wks and a trend
There were no significant group x time interaction effects for depression and anxiety, with similar
Moderate
cortisol patterns in patients with adjustment disorder and depressed mood. with depressed mood, recruited from a psychiatric outpatient clinic treatment as usual. This was delivered in 8 wkly group sessions. psycho-education
cortisol 45 min after, 12.00, 17.00, 21.00)
against time towards a steeper slope at 8 wks in the intervention (BMS) group compared to the control group.
reductions in both groups. There was no group x time interaction effect for the occurrence of suicidal ideation but there was a trend toward a greater decrease in occurrence of suicidal ideation in the BMS group compared with the control.
Krajewsk i et al. 2011 (36)
To explore the impact of different ways of spending lunch-time breaks on cortisol. 14 call centre employees Progressive muscle relaxation (PMR) performed by the participant for 20 min during
Small-talk (ST) for 20 min, with self-chosen colleagues in the staff room,
Salivary cortisol 24 wks 1 day 5 sample points (On awakening, 30 min after,
1) CAR (delta): 30 min sample minus awakening sample 2) CAR (mean): mean of awakening
There was a significant group x time interaction effect for CARdelta and bedtime cortisol, with a reduction in the PMR
N/A Moderate
lunch-break in a ‘silent room’ over a 6-month period. over a period of 6 months. 11.55, 13.05, bedtime) 11.55 and 13.05 samples were done pre and post the intervention to assess its immediate effects
plus 30 min sample 3) Pre-post intervention cortisol effect: 13.05 sample minus 11.55 sample 4) Bedtime cortisol: to assess ‘spillover’ effects Cortisol assessment over 8 timepoints
group relative to the ST group. For CAR, the change occurred towards the end of the 6 month period whereas there was a short-term and long-term effect on bedtime cortisol. Pre-posts acute changes not relevant to this review
Table 1
(continued)
Letour-neau et al. 2011 (39)To evaluate the effect of a home-based peer support intervention, that included maternal –infant interaction teaching, on mother-infant interactions and other secondary measures (Infant measures not reported in this review as not relevant) 60 mother-infant dyads in which the mother had a diagnosis of post-partum depression < 9 months after delivery
Peer support from a local volunteer, which included teaching on maternal-infant interaction techniques x 12 weeks Waiting list group x 12 weeks (received peer support after 12 weeks) 1) Mother-infant interaction (primary) 2) Maternal depressive symptoms 3) Maternal perception of social support 4) Maternal salivary cortisol 5) Infant measures (not addressed here) 12 wks 1 day 4 sample points (On awakening, noon, mid-afternoon, before bed)
AUC There were no significant time or treatment effects on maternal cortisol AUC.
There was a significant difference between groups in maternal-infant teaching interactions, favouring the control. Depressive symptoms improved significantly over time in both groups.
Moderate
Limm et al. 2011 (40)
To test the long-term effect of a stress-management intervention on self-perceived stress reactivity in employees 174 lower and middle management employees in an international plant Stress management intervention delivered over 2 full days, with booster sessions over the following 8 months Waiting-list group x 1 year 1) Perceived stress reactivity 2) Effort-reward imbalance measure 3) Anxiety and depression measures 4) Salivary cortisol 5) Salivary α-amylase 1 year 1 day 7 sample points (On awakening, 30 min, 60 min, 8am, 11am, 3pm, 8pm) 1) CAR measured in two ways: a) by subtracting the ‘awakening level’ from the 30min level and b) by calculating the AUC for the morning values 2) Diurnal decline slope measured by linear regression 3) AUC using all timepoints
There were no significant findings for the salivary cortisol parameters There was a significant improvement in self-perceived stress reactivity in the intervention group compared with the control High Lipschitz et al. 2013 (42) To explore whether salivary α-amylase (sAA) and salivary cortisol levels are positively modulated by sleep-focused mind-body interventions in cancer survivors with sleep disturbance (pilot) Non-biological outcomes reported in Nakamura et al. 2013 (58) 57 cancer survivors with self-reported sleep disturbance Mind-Body Bridging (MBB) classes delivered in once wkly 2-hour sessions over 3 wks OR Mindfulness Meditation (MM) classes delivered in once wkly 2-hour sessions over 3 wks Sleep Hygiene Education (SHE) classes delivered in once wkly 2-hour sessions over 3 wks 1) Salivary cortisol 2) Salivary sAA 3) Sleep problems 4) Perceived stress 5) Quality of life 3 wks 2 different schedules over 2 days (1 day of each schedule) 4 sample points on day 1 (30min post-awakening, noon, afternoon, evening) 1 sample point on day 2 (on awakening) 1) Waking cortisol 2) AUC: only the 4 measures from the first day included 2) Post-intervention diurnal profile modelled using mixed-effects ANCOVA: only the 4 measures from the first day included, with intervention and intervention x collection time as fixed factors
There were no significant group main effects or group x collection time effects, post- intervention, for any of the cortisol parameters.
Mean sleep problems score was significantly lower in the MBB group in comparison with the SHE group, while MBB and MM did not differ, post-intervention.
High
Lok et al. 2012 (44)
To investigate whether HPA axis activity in recurrent depressive disorder is influenced by cognitive therapy, along with other objectives relating to an embedded case-control study. Non-biological outcomes and RCT design reported in Bockting et al. 2005 (60) 187 highly recurrent major depressive disorder patients Cognitive therapy delivered in once weekly 2-hour group sessions over 8 wks
Usual care 1) Depression recurrence 2) Salivary cortisol 3) Other secondary self-report outcomes 2 years 2 different schedules over 2 days: 2 sample points on Day 1: 8am and 8pm 1 sample point on day 2: 8am Profile was modelled using mixed linear modelling with group, follow-up point and sample moment as independent variables
There was a borderline significant effect of steeper cortisol declines over the day throughout the follow-up period.
Cognitive therapy was reported to have had a significant protective effect on depression recurrence over the 2-year follow-up period in Bockting et al. 2005 (60). High Nickel C et al. 2007(46) & Nickel M.K. 2007 (47) To determine the effectiveness of behavioural/psycho-educational group training in men suffering from chronic occupational stress 72 men (≤ 65 years) who self-identified themselves as suffering from chronic occupational stress Behavioural psycho-educational group therapy delivered in twice weekly 90 min sessions over 8 weeks 90min group meetings twice weekly over 8 weeks, during which participants reported on work events 1) Systolic blood pressure 2) Salivary cortisol 3) Self-report measures of chronic stress, anger and health-related quality of life 8 wks 5 days 4 sample points (On awakening, 15, 30 and 60 min after awakening) Time-specific cortisol level (averaged over 5 days for each collection time)
There was a significant decrease in cortisol level at all collection time points in the treatment group in comparison with the control group.
There was a significant improvement in systolic blood pressure and in most of the subscales measuring chronic stress, anger and health-related quality of life in the intervention group in comparison with the control group. Moderate Lindh-Astrand et al. 2013 (41)
To study the efficacy of group therapy with applied relaxation on vasomotor symptoms in postmenopausal women 60 healthy postmenopau sal women Applied relaxation (AR) therapy delivered in 10 group sessions over 12 wks Non-treatment control 1) Average number of moderate and severe hot flashes per 24 hours (primary) 2) Quality of life 3) Salivary cortisol 24 wks 1 day 3 sample points (On awakening, 30 min after, bedtime) Time-specific cortisol levels
‘Morning cortisol’ was significantly lower in the treatment (AR) group in comparison with the control at 24 wks but not at 12 wks
There was a significant decrease in number of hot flashes in the treatment group at 12 and 24 wks in the AR group, in comparison with the control group. There was also a significant
Moderate
improvement in a number of quality of life dimensions in the AR group relative to the control.
Nunes et To examine the effects of 34 women RVT delivered Usual care 1) Psychological 24 1 day 1) Time-specific There was no significant There was a High al. 2007
(48)
relaxation and visualisation therapy (RVT) on psychological distress, cortsol levels, and immunological parameters of breast cancer patients undergoing radiotherapy with stage I or II breast cancer undergoing radiotherapy daily in group sessions over 24 days, immediately after radiotherapy distress including self-report scales of stress, anxiety and depression 2) Salivary cortisol 3) Peripheral blood mononuclear cells days 3 sample points (8am, 12 noon, 8pm) cortisol level 2) Total AUC change in cortisol parameters in either the intervention or control group
significant reduction in stress, anxiety and depression scores in the intervention group
Oken et al. 2010 (49) To evaluate whether a mindfulness meditation intervention might be effective in caregivers of close relatives with dementia and to help refine the protocol for future trials (pilot study) 31 healthy adults caring for a close relative with dementia Mindfulness-based cognitive therapy (MBCT): delivered in once wkly 90 min sessions delivered over 6 wks Dementia education classes: matched with the mindfulness meditation intervention in relation to time, social support, discussion time and home assignments OR Respite-only intervention: 3 hours wkly for 7 wks 1) Perceived stress (revised memory and behaviour problems checklist) 2) Salivary cortisol 3) Several exploratory secondary measures 7 wks 1 day 3 sample points (On awakening, 30 min after awakening, bedtime) Time-specific cortisol levels: each collection point analysed separately
There were no significant differences between groups for any of the cortisol measures post-intervention
Perceived stress was significantly lower in the MBCT and education groups in comparison to the respite group post-intervention
Moderate
Pacella et al. 2014 (50)
To examine the impact of successful PTSD treatment on the cortisol awakening response (CAR) 29 adults with chronic PTST (subsample from larger RCT) Psychotherapy delivered in once weekly 90-120 min sessions over 10 wks Sertraline titrated upwards as indicated, monitored by a psychiatrist 1) PTSD symptoms 2) Depression 3) Salivary cortisol 10 wks 1 day 4 sample points (on awakening, 30, 45 and 60 min after awakening) CAR: measured by AUCg and AUCi
Changes in CAR AUC did not differ between groups. CAR AUCi or AUCg were not predicted by clinical response to treatment in a regression analysis. 23 out of the 29 participants responded to either psychotherapy or sertraline in relation to PTSD symptoms. Moderate
Table 1
(continued)
Sears et al. 2007 (54) To investigate whether an ICD (implantable cardioverter defibrillator) shock and stress management program reduces psychological and physiological markers of anxiety and increases quality of life in patients with ICDs30 adult patients with ICDs and a history of at least one shock in the previous year Cognitive Behavioural therapy focused on shock and stress management delivered over 6 weeks 1-day psycho-educational workshop 1) Psychological measures of anxiety and depression 2) Quality of life 3) Device acceptance 4) Salivary cortisol 5) Inflammatory markers 4 wks for cortisol 1 day 6 sample points (times not reported) Mean diurnal cortisol level
There was a significant reduction in mean cortisol in the overall sample (n=30). However, there was no time x group interaction effect in relation to mean diurnal cortisol level
There was a significant improvement in anxiety, depression and quality of life in the overall sample and improvements in anxiety and mental quality of life were significantly greater the CBT group Moderate Taylor et al. 2009 (55)
To determine the effects of improving depression in depressed older patients with elevated cardiovascular risk 48 older adults (≥55 years) with depression and elevated cardiac risk Cognitive behavioural therapy (at least 10 sessions) Waiting list control 1) Depression measures 2) Cardiovascular risk factor measurement 3) Salivary cortisol 24 wks 2 days 5 sample points (On awakening, 30 min after awakening, 12 noon, 5pm, 9pm) 1) Diurnal slope 2) Waking cortisol 3) CAR (referred to as the ‘cortisol rise after waking’- no details on calculation)
There was no significant difference between groups in relation to cortisol parameters at 6 months There was a significant improvement in mood in the intervention group in comparison with the control at 6 months High Urizar and Munoz, 2011(56) To determine whether participation in a prenatal cognitive behavioural stress management programme would result in lower cortisol and self-reported stress, relative to the control, in women at high risk for depression during pregnancy Infant measures not reported here as not relevant to the review
57 pregnant (6-28 weeks gestation) women at high risk for depression Cognitive behavioural stress management programme delivered over 12 weeks, with 4 booster sessions in the post-partum period
Usual care 1) Salivary
cortisol 2) Perceived stress 3) Maternal mood 72 wks 1 day 2 sample points (Morning and evening) 1) Morning cortisol 2) Evening cortisol 3) Mean diurnal cortisol level 4) Diurnal decline: difference between morning and evening cortisol values
The mean cortisol level at 18 months was significantly lower in the intervention group in comparison with the control but there were no significant differences at 6 months Women in the intervention group experienced significantly higher perceived stress than women in the control group at 6 months but no significant difference was found at 18 months
Moderate Plag et
al. 2014 (51)
To investigate the effect of cognitive behavioural therapy, in combination with physical activity program, on salivary cortisol and α-amylase levels. Psychometric outcomes reported in Gaudlitz et al. 2015 (66) 59 adults with panic disorder Cognitive behavioural therapy (CBT) delivered in 90 min sessions over 1 month, followed by a booster session + High level endurance training on a treadmill: 30 min sessions, 3 times per week
CBT delivered in the same schedule + Low level exercise: 30 min sessions, 3 times per week over 8 wks 1) Salivary cortisol 2) Salivary α-amylase 3) Anxiety 4) Clinical global impression 28 wks 1 day 5 sample points (08am, 12noon, 4pm, 8pm, 10pm)
AUC There was a significant
group x time interaction effect for AUC in an ANCOVA at 7 months, with a significantly higher AUC in the control (low-level exercise) group.
There was a significant group x time interaction effect for the Hamilton anxiety scale, with a significantly greater improvement in the high level endurance training group at 7 months. Moderate over 8 wks Both treatments ran parallel. Richter et al. 2012 (53)
To examine the effects of a cognitive-behavioural group intervention on perceived stress and salivary cortisol levels in pregnant women 161 pregnant women (10-15 weeks gestation) with symptoms of stress, anxiety or depression Cognitive behavioural group program delivered in 8 sessions (time period not stated)
Treatment as usual 1) Salivary cortisol 2) Psychological measures including perceived stress, pregnancy-specific distress, anxiety and depression Up until 3 months post-partum 1 day 5 sample points (On awakening, 30 min after awakening, 11am, 5pm, 10pm) 1) CAR 2) Total AUC
There was a significant time x group interaction for CAR at the second time-point (post-intervention- approx. 8 weeks). Within group analysis demonstrated that there was a significant decrease in CAR in the intervention group but not in the control group.
There were no significant between-group differences for the psychological measures Moderate Wilcox et al. 2014 (57)
Wilcox et al. performed a secondary statistical analysis of the salivary cortisol data collected in the Well Elderly 2 randomised controlled trial [Clark et al. (70)]. The primary aim of this RCT was to assess the effectiveness of a lifestyle intervention on mental and physical wellbeing and cognitive functioning in older people in a community setting. 460 elderly adults aged 60-95 years. Only 379 participants agreed to provide saliva samples, with 328 providing post-intervention data. Lifestyle intervention: small group (2hours weekly) and individual sessions (1 hour x 10) led by an occupational therapist over 6 months. Non-treatment control x 6 months (provided with the intervention upon completion of the study) 1) Physical and mental wellbeing (SF-36) 2) Depression 3) Cognitive outcome variables 4) Life satisfaction index 24wks 1 day 4 sample points (On awakening, 30 min after waking, before lunch, before dinner)
CAR Two types of statistical
analysis approaches were used.
1) Conventional approach: used mean cortisol and mean CAR values 2) Alternative approach: used median cortisol and median CAR values There was no significant change in CAR pre and post intervention using both approaches.
The intervention significantly improved 5 wellbeing subscale scores as well as depression and life satisfaction scores.
Moderate
Yang et al. 2009 (59)
To compare the effects of combination therapy (psychotherapy + antidepressant) and monotherapy (antidepressant alone) on salivary cortisol levels in outpatients with major depression 65 adults attending a psychiatric outpatient clinic with a diagnosis of major depressive disorder Combination therapy (psychotherapy + antidepressant) Psychotherapy delivered in 8 sessions over 2 months and antidepressant prescribed for 4 months lunch-break in a ‘silent room’ over a 6-month period. Monotherapy (antidepressant therapy) x 4 months over a period of 6 months. 1) Salivary cortisol 2) Depression measure 16 wks 1 day 5 sample points (On awakening, 30-45 min post awakening, 12 noon, 5pm, 9pm) 1) CAR measured as the difference between morning and 30-45 min value 2) Diurnal slope measured in 2 ways: a) difference between morning and evening values b) difference between 30-45min and evening values 3) Time-specific cortisol levels
1) There was a significant decrease in the 9pm cortisol level in the combined therapy group in comparison with the monotherapy group over 4 months.
2) There was a significant difference in the diurnal slope between the two groups when measured using values from 30-45 min to evening, with a steeper slope in the combined therapy group.
Combined therapy and monotherapy both resulted in a significant improvement in depression scores, with no significant difference between groups Moderate 11.55, 13.05, bedtime) 11.55 and 13.05 samples were done pre and post the intervention to assess its immediate effects
plus 30 min sample 3) Pre-post intervention cortisol effect: 13.05 sample minus 11.55 sample 4) Bedtime cortisol: to assess ‘spillover’ effects Cortisol assessment over 8 timepoints
group relative to the ST group. For CAR, the change occurred towards the end of the 6 month period whereas there was a short-term and long-term effect on bedtime cortisol.
Pre-posts acute changes not relevant to this review
Shaded rows represent studies with agreement between clinical and cortisol findings
a
Describes the number of consecutive days of sampling and the number of samples per day per time-point
bAssessed using Gough
’s Framework
c